Major Project Theme

Introduction

The process of machine learning involves a series of steps that enable computers to learn from data and make predictions or decisions. These steps include data collection and preprocessing, feature selection and engineering, model selection and training, model evaluation, deployment, and maintenance. Each step plays a crucial role in developing effective machine-learning systems. In this paragraph, we will explore the working process of machine learning and discuss the functions of a machine learning system.

Machine Learning

Machine learning is a subfield of artificial intelligence, which is broadly defined as the capability of a machine to imitate intelligent human behaviour. Artificial intelligence systems are used to perform complex tasks in a way that is similar to how humans solve problems. Machine learning is one way to use AI. It was defined in the 1950s by AI pioneer Arthur Samuel as “the field of study that gives computers the ability to learn without explicitly being programmed.” (Brown, 2021)

(FORE, 2022)

Working Process

Machine learning works based on an algorithm which finds relationships and patterns in data and generalizes this data to predict or make decisions. It could be seen following 9 steps:

1. Data Collection: We need first to decide how we will collect the data by gathering the internal data, open-source, buying it from the vendors, or generating synthetic data. Each method has pros and cons, and in some cases, we get the data from all four methodologies.  

2. Data Preprocessing: During the data processing stage, several tasks are performed, such as selecting the appropriate features, addressing imbalanced classes, conducting feature engineering, augmenting data, and normalizing and scaling the data. To ensure reproducibility, we'll save and version the metadata, data modeling, transformation pipelines, and feature stores.   

3. Feature Selection and Engineering: During this stage, we will utilize the data gathered during the planning phase to construct and train a machine-learning model. This includes monitoring model metrics, ensuring scalability and resilience, and optimizing storage and compute resources. 

• Build effective model architecture by doing extensive research.
• Defining model metrics. 
• Training and validating the model on the training and validation dataset. 
• Tracking experiments, metadata, features, code changes, and machine learning pipelines.
• Performing model compression and ensembling.
• Interpreting the results by incorporating domain knowledge experts. 

Our focus will be on model architecture, code quality, conducting machine learning experiments, training models, and implementing ensembling techniques (Awan, 2022). 

4. Model Selection: When exploring algorithms, it's important to consider your use case. By identifying your goal, you can narrow down your search for solutions. There are several methods available, such as regression, classification, clustering, recommendations, and anomaly detection, but it's important to choose the one that best fits your needs (IBM, 2023). 

5. Model Training: Learning good values for all the weights and biases from labelled examples is the essence of training a model. 

6. Model Evaluation: Before implementing our model, we will conduct a test on a separate dataset. Additionally, we will collaborate with subject matter experts to properly identify any errors in our predictions. We also need to ensure that we follow industrial, ethical, and legal frameworks for building AI solutions. Additionally, we plan to assess the robustness of our model by testing it on both randomized and real-world data. We aim to ensure that the model can provide quick and valuable inferences. At last, we will evaluate the outcomes against the pre-determined success criteria and make a decision on whether or not to implement the model. During this stage, each step is documented and kept in versions to ensure consistency and the ability to reproduce the process. 

7. Model Deployment: In this phase, we deploy machine learning models to the current system. During the deployment process, it is important to consider the inference hardware. Sufficient RAM, storage, and computing power are required to ensure speedy results. Once this is determined, we will assess the model's performance in production through A/B testing to ensure user satisfaction. The deployment strategy is important. You need to make sure that the changes are seamless and that they have improved the user experience.  

8. Model Maintenance: Once the model has been deployed to production, it is important to continually monitor and enhance the system. Our focus will be on tracking model metrics, evaluating hardware and software performance, and ensuring customer satisfaction (Awan, 2022). 

Functions of Machine Learning System

• Descriptive - the system uses the data to explain what happened.
• Predictive - the system uses the data to predict what will happen.
• Prescriptive - the system will use the data to suggest what action to take. (Brown, 2021)

Types of Machine Learning:

(Data Basecamp, 2023)

• Supervised learning: The dataset being used has been pre-labelled and classified by users to allow the algorithm to see how accurate its performance is.

• Unsupervised learning: The raw dataset being used is unlabeled and an algorithm identifies patterns and relationships within the data without help from users.

• Semi-supervised learning: The dataset contains structured and unstructured data, which guides the algorithm on its way to making independent conclusions. The combination of the two data types in one training dataset allows machine learning algorithms to learn to label unlabeled data.

• Reinforcement learning: The dataset uses a “rewards/punishments” system, offering feedback to the algorithm to learn from its own experiences by trial and error. (UCB-UMT, 2022)

Application of Machine Learning

• Speech Recognition: Automatic speech recognition (ASR), computer speech recognition, or speech-to-text are different terms used to describe the ability to convert spoken language into written text. This technology relies on natural language processing (NLP) to perform this task. Many mobile devices utilize speech recognition to enable voice search, such as Siri, and improve accessibility for text messaging.

• Customer Service: Online chatbots are replacing human agents along the customer journey, changing the way we think about customer engagement across websites and social media platforms. Chatbots answer frequently asked questions (FAQs) about topics such as shipping, or provide personalized advice, cross-selling products or suggesting sizes for users. Examples include virtual agents on e-commerce sites; messaging bots, using Slack and Facebook Messenger.

• Recommendation Engines: By analyzing past consumer behaviour data, AI algorithms can identify patterns that can be utilized to create improved cross-selling tactics. Online retailers use this approach to suggest relevant products to customers at checkout.

• Automated Stock Trading: Designed to optimize stock portfolios, AI-driven high-frequency trading platforms make thousands or even millions of trades per day without human intervention.

• Fraud Detection: Financial institutions, including banks, can leverage machine learning to detect suspicious transactions. By utilizing supervised learning, a model can be trained with data on previously identified fraudulent transactions. Additionally, anomaly detection can flag transactions that deviate from typical patterns and may require further investigation (IBM, 2023).

Conclusion

Machine learning has revolutionized the field of artificial intelligence by providing systems with the ability to learn from data and improve their performance over time. Through a series of steps, including data collection, preprocessing, model selection, training, evaluation, deployment, and maintenance, machine learning systems can extract patterns and relationships from data and make informed predictions or decisions. These systems find applications in various domains, including speech recognition, customer service, recommendation engines, automated stock trading, and fraud detection. With further advancements in machine learning techniques and technologies, the potential for its application and impact will only continue to grow.

Reference list:

Brown, S. (2021) Machine Learning, explained, MIT Sloan. Available at: https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained (Accessed: April 12, 2023).

Data Basecamp (2023) Reinforcement learning - simply explained!. Available at: https://databasecamp.de/en/ml/reinforcement-learnings (Accessed: April 13, 2023).

FORE School Of Management (2022) What is machine learning course: Its importance and types-fore. Available at: https://www.fsm.ac.in/blog/an-introduction-to-machine-learning-its-importance-types-and-applications/ (Accessed: April 12, 2023).

UCB-UMT  (2022) What is machine learning (ML)?.  Available at: https://ischoolonline.berkeley.edu/blog/what-is-machine-learning/ (Accessed: April 13, 2023).

Awan, A.A. (2022) The Machine Learning Life Cycle explained, DataCamp. Available at: https://www.datacamp.com/machine-learning-lifecycle-explained (Accessed: 01 June 2023).

IBM  (2023) Evaluate and select a machine learning algorithm. Available at: https://www.ibm.com/garage/method/practices/reason/evaluate-and-select-machine-learning-algorithm/ (Accessed: 01 June 2023).

IBM (2023) What is machine learning?. Available at: https://www.ibm.com/topics/machine-learning (Accessed: 01 June 2023).